On Mon, 2009-10-05 at 00:15 +0100, Antonio Marcos López Alonso wrote: > > Is this a new problem or did it occur with earlier kernel versions? > > No, it happened also in previous versions. But at least irqpoll/irqfixup > worked pretty well. Now this behavior seems to get worsened even using these > kernel options. > > > > > Can you try to reproduce this without the nvidia or virtualbox modules > > loaded? > > I can but just to make things faster: > > Jens-Michael, > > Have you got any nvidia/virtualbox modules running in your host? Just to > discard... The warning message shows all loaded modules, and those aren't included, so this question is answered. I had a look at the code and the values in the 'transmit timed out' message, and it seems that the NIC has reported a transmit completion but this hasn't been handled. Perhaps another device sharing its IRQ is misbehaving and causing the IRQ to be disabled. Please can you send more of the kernel log from before the TX watchdog warning? Also, if this happens again, please send the contents of /proc/interrupts. Ben. -- Ben Hutchings I say we take off; nuke the site from orbit. It's the only way to be sure.
Attachment:
signature.asc
Description: This is a digitally signed message part